# Web Science & Technologies University of Koblenz ▪ Landau, Germany Advanced Data Modeling Minimal Models Steffen Staab TexPoint fonts used in EMF. Read.

## Presentation on theme: "Web Science & Technologies University of Koblenz ▪ Landau, Germany Advanced Data Modeling Minimal Models Steffen Staab TexPoint fonts used in EMF. Read."— Presentation transcript:

Web Science & Technologies University of Koblenz ▪ Landau, Germany Advanced Data Modeling Minimal Models Steffen Staab TexPoint fonts used in EMF. Read the TexPoint manual before you delete this box.: A A AAA

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 2 of 37 WeST Overview  Logic as query language.  Grounding.  Minimal Herbrand models.  Completion.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 3 of 37 WeST Logic as query language Given:  first-order formula A[x 1, …, x n ]  Herbrand interpretation I  This first-order formula can be considered as a definition of a relation R A on T n § as follows:  (t 1, …, t n ) 2 R A := I ² A[t 1, …, t n ]

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 4 of 37 WeST Clauses as definitions We say that a clause p(t 1, …, t m ) :- L 1, …, L n defines the relation symbol p. Let C be a set of clauses and p be a relation symbol. We call the definition of p in C the set of all clauses in C that define p.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 5 of 37 WeST Principles of semantics A deductive database is a set of clauses. This set of clauses is regarded as a collection of definitions of relations. The semantics defines the meaning of this definitions by associating with them an interpretation, or a class of interpretations. Query answering is based on the semantics.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 6 of 37 WeST Two key assumptions  the unique name assumption: each name denotes a unique object.  the closed world assumption:  a negative statement : A holds if the corresponding positive one A does not hold. Both assumptions are not supported by (Tarskian) models for first-order logics (why not?)  One solution: minimal models

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 7 of 37 WeST Minimal Herbrand Models Let I be a Herbrand model of a set of formulas S. We call I a minimal Herbrand model of S if it is minimal w.r.t. the subset relation, i.e. for every Herbrand model I´of S of the same signature we have I´ ¶ I. I is called the least Herbrand model of S if for every Herbrand model I´ of S of the same signature we have I µ I´.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 8 of 37 WeST Does every set of formulas S have a least Herbrand model?

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 9 of 37 WeST Does every set of formulas S have a least Herbrand model? Counterexample for normal clauses: person(a). man(X) :- person(X), not woman(X). woman(X) :- person(X), not man(X).

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 10 of 37 WeST Does every set of formulas S have a least Herbrand model? What about definite clauses?

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 11 of 37 WeST Instance and variant Let E, E´ be a pair of terms or formulas. E´ is an instance of E, denoted E > E´, if there exists a substitution µ such that E µ = E´. ground instance: instance that is ground, E´ is a variant of E if E´ is an instance of E and E is an instance of E´.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 12 of 37 WeST Examples  P(x,a) is instance of P(x,y) because of P(x,y)[y|a]  P(b,a) is a ground instance  P(x,y) and P(u,v) are variants of each other, because of  [x|u, y|v] and  [u|x, v|y]

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 13 of 37 WeST Grounding  Let C be a set of clauses and § be any signature containing all symbols used in C. The grounding of C w.r.t. §, denoted C * is the set of all ground instances of the signature § of clauses in C.  Lemma. Let I be a Herbrand interpretation and C be a set of clauses. Then I ² C if and only if I ² C *.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 14 of 37 WeST General proof scheme First order formula (all possible ways of) grounding Propositional formula Propositional Consequences First order Consequences lifting calculus

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 15 of 37 WeST  Proof

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 16 of 37 WeST Logic with equality  Additional atomic formulas s = t, where s, t are terms.  Abbreviation: x  y := : (x = y).  Unlike other relations, the semantics of s = t is predefined in all Herbrand interpretations: I ² s = t if s coincides with t.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 17 of 37 WeST Example valid formulas  f(x 1, …, x n ) = f(y 1, … y n ) ! x 1 = y 1 Æ … Æ x n = y n  f(x 1, …, x n )  g(y 1, …, y n )  f(x 1, …,x n )  c  d  c  A[t] \$ 8 x(x = t ! A[x])

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 18 of 37 WeST Semantics of definitions Consider a definition of a relation r What is the meaning of this definition? Is there a largest Herbrand model? Especially in the light of negation as failure?

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 19 of 37 WeST Idea of completion man(hans). man(adam). person(eva). woman(X) :- person(X), not man(X). Is eva a woman? She might be a man and we just don‘t know! Minimal model says that she is not a woman! Trick: Completion: man(X) \$ (X=hans Ç X=adam). woman(X) \$ X=eva. I.e. hans and adam are the only men.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 20 of 37 WeST Idea of Completion man(hans). man(X) :- lovesBeer(X,Y). Completion: man(X) \$ X=hans Ç ( 9 Y: X=U Æ lovesBeer(U,Y)) A man is a man only if he is hans or if he loves some brand of beer.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 21 of 37 WeST Completion. Step 1. Replace every clause by an equivalent one such that the arguments of r are x 1, …, x n : Given: r(t 1, …, t n ) :- G Replace by: r(x 1, …, x n ) :- x 1 = t 1 Æ … Æ x n = t n Æ G

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 22 of 37 WeST Completion. Step 2. If there are variables y 1, …, y k occurring in a body but not in the head, apply 9 to these variables, i.e., Given r(x 1, …, x n ) :- G Modify to r(x 1, …, x n ) :- 9 y 1 … 9 y k G

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 23 of 37 WeST Completion. Step 3. If there are several definitions, replace them by one Given r(x 1, …,x n ) :- G 1 … r(x 1, …, x n ) :- G m Replace by r(x 1, …, x n ) :- G 1 Ç … Ç G m

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 24 of 37 WeST Completion. Step 4. Replace :- by \$ : Given r(x 1, …, x n ) :- G 1 Ç … Ç G m Replace by r(x 1, …, x n ) \$ G 1 Ç … Ç G m The formula r(x 1, …, x n ) \$ G 1 Ç … Ç G m is called the completed definition of the original set of clauses.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 25 of 37 WeST Example r(u,v):-p(u,1,z). r(v,u):-p(2,u,z). r(x,y) :- x=u Æ y=v Æ p(u,1,z). r(x,y) :- x=v Æ y=u Æ p(2,v,z). r(x,y) :- 9 z,u,v x=u Æ y=v Æ p(u,1,z). r(x,y) :- 9 z,u,v x=u Æ y=v Æ p(2,v,z). r(x,y) \$ ( 9 z,u,v x=u Æ y=v Æ p(u,1,z)) Ç ( 9 z,u,v x=u Æ y=v Æ p(2,v,z))

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 26 of 37 WeST Properties  All steps preserve Herbrand models, except for the last one.  Why?  Gives a unique semantics to non-recursive definitions  What about recursive definitions?  Logic programming semantics and first-order semantics coincide for definite programs

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 27 of 37 WeST Recursive definitions odd(1). even(f(X)) :- odd(X). odd(f(X)) :- even(X). Completion: odd(X) \$ X=1 Ç (X=f(Y) Æ even(Y)). even(X) \$ X=f(Y) Æ odd(Y).

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 28 of 37 WeST Recursive definitions person(adam). person(eva). woman(X) :- person(X), not man(X). man(X) :- person(X), not woman(X). Completion: woman(X) \$ (Y=X Æ person(Y) Æ not man(Y)). man(X) \$ (Y=X Æ person(Y) Æ not woman(Y)). Semantics not unique in logic programming: Models are I={woman(adam),woman(eva)} I={man(adam),man(eva)} I={woman(adam),man(eva)} I={man(adam),woman(eva)} What is the semantics in first order logics?

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 29 of 37 WeST Recursive definitions person(adam). person(eva). woman(X) :- person(X), not man(X). man(X) :- person(X), not woman(X). Completion: woman(X) \$ (Y=X Æ person(Y) Æ not man(Y)). man(X) \$ (Y=X Æ person(Y) Æ not woman(Y)). Semantics not unique in logic programming: Models are (add {person(adam),person(eva)} to each) I={woman(adam),woman(eva)} I={man(adam),man(eva)} I={woman(adam),man(eva)} I={man(adam),woman(eva)} What is the semantics in first order logics? Models are: woman I ={}=man I

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 30 of 37 WeST Simple characterization of completion Let C be a definition of r, I be a Herbrand model of the corresponding completed definition, and r(t 1, …, t n ) be a ground atom. Then I ² r(t 1, …, t n ), 9 (r(t 1, …, t n ) :- L 1, …, L m ) 2 C * (I ² L 1 Æ … Æ L n ):

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 31 of 37 WeST Immediate consequence operator T C (I) := { A | there exists (A :- G) 2 C * such that I ² G } Fixpoint: an interpretation such that T C (I) = I.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 32 of 37 WeST Definite clauses have the least model Let C be a set of definite clauses. Define I 0 := {} I n+1 := T C (I n ), for all n ¸ 0, I ! := [ ! i=0 I i Then I ! is the least fixpoint of T C and also the least Herbrand model of C.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 33 of 37 WeST Non-recursive databases Let C be a non-recursive database and K be an arbitrary interpretation. Define I 0 := K I n+1 := T C (I n ), for all n ¸ 0, I ! := [ ! i=0 I i Then I ! is the only fixpoint of T C. Moreover, for some n we have I ! = I n.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 34 of 37 WeST Non-recursive sets of clauses  Let C be a set of clauses.  Its dependency graph consists of pairs p ! r such that p occurs in the body of a clause which defines r in C.  A set of clauses is non-recursive if the dependency graph contains no cycles.

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 35 of 37 WeST Non-recursive sets of clauses Let C be a set of clauses. Its dependency graph consists of pairs p ! r such that p occurs in the body of a clause which defines r in C. A set of clauses is non-recursive if the dependency graph contains no cycles. Person(a). Woman(X) :- Person(X), Not Man(X). Man(X) :- Person(X), Not Woman(X). WomanMan Person

Steffen Staab staab@uni-koblenz.de Advanced Data Modeling 36 of 37 WeST Non-recursive sets of clauses Dependency graph of C consists of pairs p ! r such that p occurs in the body of a clause which defines r in C. C is non-recursive if the dependency graph contains no cycles. A relation p depends on a relation q in C if there exists a path of length ¸ 1 from q to p in the dependency graph of C. A set of clauses is non-recursive if and only if no relation depends on itself.

Download ppt "Web Science & Technologies University of Koblenz ▪ Landau, Germany Advanced Data Modeling Minimal Models Steffen Staab TexPoint fonts used in EMF. Read."

Similar presentations